[syzbot] [net?] WARNING in qdisc_tree_reduce_backlog (2)

[syzbot]

Hello,

syzbot found the following issue on:

HEAD commit: 5862221fdded Merge tag 'parisc-for-7.1-rc3' of git://git.k..
git tree: upstream
console output: https://syzkaller.appspot.com/x/log.txt?x=12a4bb26580000
kernel config: https://syzkaller.appspot.com/x/.config?x=f2e8ebfec4636d32
dashboard link: https://syzkaller.appspot.com/bug?extid=9744ccaabe337c6fb123
compiler: Debian clang version 21.1.8 (++20251221033036+2078da43e25a-1~exp1~20251221153213.50), Debian LLD 21.1.8

Unfortunately, I don't have any reproducer for this issue yet.

Downloadable assets:
disk image: https://storage.googleapis.com/syzbot-assets/cd9aba7e59bf/disk-5862221f.raw.xz
vmlinux: https://storage.googleapis.com/syzbot-assets/29af9d57e9af/vmlinux-5862221f.xz
kernel image: https://storage.googleapis.com/syzbot-assets/02749594fd1e/bzImage-5862221f.xz

IMPORTANT: if you fix the issue, please add the following tag to the commit:
Reported-by: syzbot+9744cc...@syzkaller.appspotmail.com

------------[ cut here ]------------
parentid != TC_H_ROOT
WARNING: net/sched/sch_api.c:797 at qdisc_tree_reduce_backlog+0x3d9/0x480 net/sched/sch_api.c:797, CPU#1: ktimers/1/29
Modules linked in:
CPU: 1 UID: 0 PID: 29 Comm: ktimers/1 Tainted: G L syzkaller #0 PREEMPT_{RT,(full)}
Tainted: [L]=SOFTLOCKUP
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 04/18/2026
RIP: 0010:qdisc_tree_reduce_backlog+0x3d9/0x480 net/sched/sch_api.c:797
Code: ff ff 4c 89 ef e8 b7 85 12 f9 e9 42 ff ff ff e8 4d 7c ab f8 eb 17 e8 46 7c ab f8 eb 10 e8 3f 7c ab f8 eb 09 e8 38 7c ab f8 90 <0f> 0b 90 e8 7f 72 03 02 89 c3 31 ff 89 c6 e8 d4 80 ab f8 85 db 74
RSP: 0018:ffffc90000a3f768 EFLAGS: 00010246
RAX: ffffffff8918f818 RBX: 0000000000000008 RCX: ffff88801daa3d80
RDX: 0000000000000100 RSI: 0000000000000000 RDI: 0000000000000100
RBP: dffffc0000000000 R08: 0000000000000000 R09: 0000000000000100
R10: 0000000000000100 R11: 00000000ffffffff R12: 00000000000affe0
R13: dffffc0000000000 R14: ffffc90000a3f8e0 R15: ffff88803d0a7800
FS: 0000000000000000(0000) GS:ffff888126279000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 000055558bcbda38 CR3: 00000000403d8000 CR4: 00000000003526f0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000003e4f
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Call Trace:
<TASK>
sfq_rehash net/sched/sch_sfq.c:598 [inline]
sfq_perturbation+0x205d/0x22d0 net/sched/sch_sfq.c:615
call_timer_fn+0x192/0x5e0 kernel/time/timer.c:1748
expire_timers kernel/time/timer.c:1799 [inline]
__run_timers kernel/time/timer.c:2374 [inline]
__run_timer_base+0x6a3/0x9f0 kernel/time/timer.c:2386
run_timer_base kernel/time/timer.c:2395 [inline]
run_timer_softirq+0x103/0x170 kernel/time/timer.c:2406
handle_softirqs+0x1de/0x6d0 kernel/softirq.c:622
__do_softirq kernel/softirq.c:656 [inline]
run_ktimerd+0x69/0x100 kernel/softirq.c:1151
smpboot_thread_fn+0x541/0xa50 kernel/smpboot.c:160
kthread+0x388/0x470 kernel/kthread.c:436
ret_from_fork+0x514/0xb70 arch/x86/kernel/process.c:158
ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:245
</TASK>


---
This report is generated by a bot. It may contain errors.
See https://goo.gl/tpsmEJ for more information about syzbot.
syzbot engineers can be reached at syzk...@googlegroups.com.

syzbot will keep track of this issue. See:
https://goo.gl/tpsmEJ#status for how to communicate with syzbot.

If the report is already addressed, let syzbot know by replying with:
#syz fix: exact-commit-title

If you want to overwrite report's subsystems, reply with:
#syz set subsystems: new-subsystem
(See the list of subsystem names on the web dashboard)

If the report is a duplicate of another one, reply with:
#syz dup: exact-subject-of-another-report

If you want to undo deduplication, reply with:
#syz undup

[Jiayuan Chen]

I think the issue is that before commit 47e8dbb6e763 ("net/sched: do not
reset queues in graft operations"),
dev_deactivate() reset the per-tx-queue of lower leaf qdiscs (including
any sfq) before dev->qdisc was swapped.

After 47e8dbb6e763, dev_deactivate(dev, false) skips that reset. The
leaf will be drained much later, inside __qdisc_destroy(leaf).
But the timer sfq_perturbation may be fired between
rcu_assign_pointer(dev->qdisc, new) and __qdisc_destroy, and dev->qdisc
already
points at the new root.

May be the simplest way is adding test_bit(__QDISC_STATE_DEACTIVATED,
&sch->state) at the start of sfq_perturbation.